To our knowledge there are no effective pharmacologic treatments for marijuana dependence. One subgroup that might benefit from a targeted pharmacologic intervention is marijuana-dependent adolescents with attention-deficit/hyperactivity disorder (ADHD). The estimated prevalence of ADHD in substance-abusing adolescents in clinical samples ranges from 20-35%. This rate is substantially higher than what is found in the general population (2-10%). Given the comorbidity of depression and marijuana dependence, pharmacotherapies aimed at treating these dually-diagnosed patients may be particularly effective. Atomoxetine, a recently FDA-approved nonstimulant medication for child and adult ADHD, with low abuse liability, is a promising option for this patient population. Recent work completed by our group has found that individuals with ADHD are less likely to complete substance abuse treatment than those without ADHD. Further, the literature suggests that adolescents with ADHD are less likely to attain treatment goals than those without ADHD. It is becoming increasingly clear that no one intervention, whether it be pharmacologic or nonpharmacologic, will be successful in treating substance-dependent individuals with comorbid ADHD. Instead, an integrated treatment is needed. We therefore propose a placebo-controlled trial of atomoxetine, in a sample of marijuana-dependent adolescents with ADHD. The study will be conducted at Phoenix House's adolescent therapeutic community in Shrub Oaks, New York. All participants in the trial will also receive cognitive skills training. The following specific aims will be addressed: 1. To determine whether atomoxetine reduces symptoms of ADHD compared to placebo. 2. To determine whether atomoxetine improves treatment retention compared to placebo. 3. To determine whether atomoxetine improves progress in treatment compared to placebo. 4. To examine the relationship between medication treatment, reduction in ADHD symptoms, and improvement in retention and/or progress in treatment using mediational analyses.